Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Analytics › Troubleshooting

GA4 Direct Traffic Too High

Why Google Analytics shows too much direct traffic: untagged email and app links, redirects that strip parameters, lost referrers and cross-domain gaps.
Troubleshootingintermediate6 min readUpdated

First, confirm direct is inflated, and locate it

Start with Traffic acquisition filtered to the Direct channel, then switch the primary dimension to Landing page, and add device category as a secondary dimension. Read the list as a map of leaks. The home page and short, memorable paths are where genuine direct visits arrive. Long product URLs, blog posts and campaign landing pages with large direct entry are not typed by anyone; something sent those visitors there and the evidence was lost. A direct share far higher on mobile points at apps and messaging, which often open links without a referrer. A direct spike on the day a newsletter goes out points at untagged email links.

Compare the period with the same landing-page view a few months earlier, because a step change dates the leak and usually matches a release, a new email platform or a site migration. Two related symptoms belong elsewhere: sessions under "(not set)" or a broken medium are covered by the sibling page on GA4 unassigned traffic, and the general case of campaign data lost on the way in has its own short section in common GA4 mistakes. This page takes the landing-page view and works through each leak in turn.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Direct spikes on send days; deep pages linked from email, SMS or documents show direct entriesClick the owned links and inspect the landing URL for campaign parametersUntagged links in email, SMS, PDFs and appsTag every owned link from the convention's builder
Campaign landing pages show direct while the ads report clicksFollow the ad or short link and watch the redirect chain in the network panelRedirects that drop UTM parameters or click identifiersPreserve the full query string through every redirect
Direct entries on pages linked from partner or older sitesThe linking page's protocol and referrer policyReferrer lost between secure and insecure pages or by referrer policyServe the site securely; ask partners to tag links
Direct sessions start on checkout, booking or second-domain pagesWhether the linker parameter appears on the cross-domain hopCross-domain journeys not configuredList every domain in the stream's cross-domain setting
Direct on odd pages at odd hours with almost no engagementHour, location, engagement time and landing page of the direct sessionsAutomated traffic arriving with no referrerSeparate and exclude bots; investigate the source

Fix, verify and prevent each leak

Email platforms, SMS tools, PDF brochures, QR codes, app push notifications and social bios all send visitors without a referrer that names them. Without campaign parameters, every one of those visits is direct.

Fix: tag every link the organisation controls with source, medium and campaign from the shared builder in the UTM and naming conventions guide.

Verify: click each corrected link from a clean browser and watch Realtime record the intended source and medium.

Prevent: make "links generated from the builder and tested" a line on every send and launch checklist, and include templates and automated emails, which are tagged once and forgotten.

Redirects that strip the evidence

Link shorteners, vanity domains, tracking redirects, a forced move to a trailing-slash URL or a consent page can each drop the query string, taking the tags and any click identifier with them. The visit then lands with nothing and becomes direct.

Fix: configure every redirect to pass the full query string to the final URL; the marketing data quality guide treats parameter-stripping redirects as a standing failure to test for.

Verify: follow a tagged link in the network panel and confirm the parameters survive each hop to the final page.

Prevent: add a redirect check to the release process for any change to routing, domains or the consent flow.

Referrers the browser did not send

A visit carries its origin in the referrer only if the linking page allows it. Links from an insecure page to a secure one, linking sites with a strict referrer policy, and many in-app browsers send no referrer, so the visit arrives as direct.

Fix: serve every page securely, keep your own referrer policy from hiding internal origins, and ask the partners and publishers that send meaningful traffic to tag their links, as the data quality guide recommends for consistent sourcing.

Verify: a test click from the partner page now arrives with its source.

Prevent: keep a list of high-value referring partners and give each one tagged links during onboarding.

Domain hops that start a new visit

When a journey moves from the main site to a booking engine, shop or help centre on another domain without cross-domain measurement, the second domain cannot recognise the visitor. Depending on how the hop is made and which domains are excluded as referrals, the new session can appear as a self-referral or as direct.

Fix: list every domain in the stream's cross-domain setting and confirm the linker parameter is added, as the GA4 setup checklist describes.

Verify: the same client identifier appears on both sides of the hop in DebugView, and direct entries on the second domain fall.

Prevent: add any new domain to the list before it goes live, not after reports show the split. When the booking engine or widget belongs to a vendor that does not carry your tag, the guide to tracking embedded widgets and hosted pages shows how to record the outcome on your own domain and keep its source.

Automated visits

Scripts and crawlers that slip past known-bot exclusion usually send no referrer. They show as direct, often to unusual pages, at unusual hours, with almost no engagement.

Fix: isolate them with the landing-page, hour and location view, then follow the sibling page on a sudden traffic spike to decide between filtering and blocking.

Verify: direct's engagement rate returns towards its usual level once the automated sessions are separated.

Prevent: watch direct's engagement alongside its volume; a jump in volume with a fall in engagement is the bot signature.

Direct can only be reduced by leaving fewer visits without evidence, and most of that evidence is under the organisation's control. Inventory every outbound channel that sends people to the site: newsletters, lifecycle emails, SMS, app notifications, documents, QR codes, social profiles, partner placements. Video descriptions belong on the list too, since clicks from the YouTube apps often arrive without a referrer, and the guide to measuring YouTube's contribution to the business shows what tagged and untagged YouTube visits can each prove. Assign each a source and medium from the convention and generate its links from the builder. Test each new link once, end to end, including any redirect. Review direct by landing page monthly and treat a deep page with heavy direct entry as a ticket. Keep the cross-domain list in step with the domains the journey touches. Remember why this matters beyond tidiness: channel reports and every Attribution view built on sessions understate whichever channel's links went untagged, and budgets follow those reports.

Illustrative example — not a real result

A training provider's direct traffic, located by landing page

A training provider sees direct at 41% of sessions, up from 24% a year earlier. Direct by landing page shows the home page in first place, as expected, but places 2 to 6 are course pages with long URLs, each with direct shares above 60%, concentrated on Tuesdays and on mobile. Tuesday is newsletter day: a change of email platform in the spring had dropped the template's automatic tagging. A second leak appears on the booking domain, where direct sessions begin at the payment step because the domain was never listed for cross-domain measurement. After retagging the newsletter template and adding the booking domain, direct falls to 27% the following month, and Email rises from 6% to 15% of sessions.

Common mistakes

Treating direct as brand demand without looking at landing pages. Tagging campaigns but not lifecycle emails and documents. Trusting a short link without following its redirect. Excluding your own domains as referrals without setting up cross-domain measurement. Reading a fix as retroactive.

Frequently asked questions

How much direct traffic is normal?

There is no universal benchmark, because direct depends on brand strength, the share of returning visitors and how much of the site's promotion runs through untagged channels. Judge direct against its own history and by where it lands. Direct sessions entering on the home page are plausible; direct sessions entering on deep article or product pages that nobody would type are the sign of lost sources.

Does direct traffic get credit for conversions?

In the attribution reports, the help centre documents that direct visits receive credit only when a path contains nothing but direct visits. In session-scoped acquisition reports, though, every direct session is shown as direct, so a large untagged channel inflates direct there and understates the channel that actually sent the visit.

Will fixing tagging reduce direct in past reports?

No. Stored session sources are not rewritten, so the fix applies from the day it ships. Annotate the date, and compare direct by landing page before and after the change to confirm the leak closed.

Terms used on this page

Where this page is used

Also related

Sources

  1. [GA4] URL builders: Collect campaign data with custom URLs — Google Analytics Help
  2. Google Analytics Help — Google Analytics Help

Related